Private & Confidential

MS Frontier Re Modeling Research Pte. Ltd.Catastrophic Risk – A Flood PerspectiveKunal Jadhav

12 April 2012

Private & Confidential 2

• Overview of natural disasters

• Flood risk assessment methodology

• Uncertainty in flood modeling

• Flood Profile – Asia & Pakistan

• Summary

Agenda

Private & Confidential 3

Overview of natural disaster

Annual Average Damage ($US billion): 1990-2010

Source:"EM-DAT: The OFDA/CRED International Disaster Database, Universite catholique de Louvain, Brussels, Belgium“

Private & Confidential 4

Economic Losses due to Flood – Asia 1965 - 2004

Private & Confidential 5

Why are the numbers are bigger now?

Economy Insurance Penetration

(source www.treasury.gov.au)

Population

Private & Confidential 6

AIR EQECAT RMS Risk Frontiers

Country Tropical Cyclone Earthquake Tropical

Cyclone EarthquakeTyphoon rainfall – flooding

Tropical Cyclone Earthquake Riverine

Flood Hail

China a a a a a   aGuam         a aHong Kong* a   a a a aIndia     a a   aIndonesia   a   a   aJapan a a a a a a aMacau*           aMalaysia     a a a    Pakistan     a a    Philippines a a a a a   aSaipan**           aSingapore       a    South Korea a    a a a    Taiwan a a a a a   aThailand     a a a    Australia a a a a a a a aNew Zealand a a a* Included in China EQ          ** Included in Guam EQ          

What is available?

Private & Confidential 7

Flood risk assessment methodology

Inundation Model • Hydrological• Hydraulic

Inundation Model • Hydrological• Hydraulic

Calibration• Reconstruction of Historical Events

Calibration• Reconstruction of Historical Events Satellite Flood ImagesSatellite Flood Images

Stochastic Events Stochastic Events

Flood Extent and Depth Flood Extent and Depth

GIS Input • DEM• River

• Cross-sections• Roughness

GIS Input • DEM• River

• Cross-sections• Roughness

Exposures at Risk• Building attributed• Policy Information

Exposures at Risk• Building attributed• Policy Information

Flood Risk Assessment Flood Risk Assessment

Vulnerability Vulnerability

Financial ModelFinancial Model

Probabilistic Risk Model Rainfall/Runoff based

Private & Confidential 8

Hydrology

Private & Confidential 9

Stochastic Module

The stochastic event module generates stochastic events from the characteristics of historical events using simulation techniques.• Duration• Intensity• Spatial pattern • Initial loss

Return Period Rainfall in mm

Private & Confidential 10

Hazard Module – Hydrologic modeling

Input for hydrological model• Digital Elevation model – Ground surface elevation

data• River Network• Soil Map• Land-use• Rainfall• Observed flow data

Private & Confidential 11

Hazard Module – Hydraulic modeling

http://www.riskscape.org.nz/structure/modelling

Input for hydraulic model• River cross section• Flow hydrograph• Surface roughness • Satellite imagery

Private & Confidential 12

Vulnerability Module

The vulnerability is the expected degree of loss, which can be expressed as a damage ratio.

• magnitude of the hazard (e.g. flood depth, flow velocity) &

• the characteristics (attributes) of the asset (e.g. structural strength of a building)

Flood Depth

Mean

Dam

ag

e R

ati

o

Physical Characteristics

Construction Type 1

Construction Type 2

Construction Type 3

Private & Confidential 13

Exposure Module

Exposure data development• LOB

• Residential• Commercial• Industrial• Agriculturaletc.

• Structural details• Construction type and material • Wall and flooring material etc• Age and heightetc.

• Location

• Data Resolution

Source: India : Census of India 2001

Private & Confidential 14

Loss Module

Source : http://www.adrc.asia/publications/Venten/HP/Paper(Herath).htm

Loss = MDR * Replacement cost

Private & Confidential 15

Source : RMSI, 2009

Flood Hazard Map

Private & Confidential 16

Modeling Uncertainties

• Hazard

• Exposure

• Vulnerability

Variation depends on mean damage ratio

Residential Wood Frame

Private & Confidential 17

Other challenges in flood modeling

Data• Meteorological and hydrological data - availability and gaps

• Large amount of data – heavy to process

• Resolution of data – Model Input Data

• Exposure data - resolution

• Loss data for historical events

Private & Confidential 18

- Building Attributes• Construction• Occupancy• Building Height• Year Built. etc

- Exact Location• Address Information

Other challenges in flood modeling

Private & Confidential 19

Flood Profile - Asia

http://www.preventionweb.net/english/hazards/statistics/risk.php?hid=62¹: In billions US$

Private & Confidential 20

Flood Profile - Asia

Private & Confidential 21

Flood Profile - Pakistan

Private & Confidential 22

• Asia’s potential for large catastrophe losses is increasing due to

– growing population

– economic growth

– increasing insurance penetration

Summary

• Currently no standalone Flood Model available in Asia

• Flood models are the most complex models – large amounts of high

resolution datasets are needed

• Data quality : Garbage in – garbage out

Private & Confidential 23

Private & Confidential 24

Types of Floods

Type Cause Areas Affected Loss AgentsDamage - Frequency and Severity

Loss Prevention

Storm Surge High level of water due to wind set-up, high waves

-Coastal areas-Areas near big lakes

-Wave force-Salt water

-Low frequency-Very high losses

-Early warning-Dykes-Evacuation

River Flooding

Intense and/or persistent rain for longer period Areas near river -Prolonged impact of water

-Water contamination-Low frequency-High loss potential

-Early Warning-Structural flood control-Temporary protection of property-Putting movable objects somewhere elseEvacuation

Flash Flood Intense rainfall; mostly locally Practically anywhere

-Mechanical impact of fast flowing water-Large amount of sediments

-High frequencyRelatively minor losses

-Adequate drainage

Private & Confidential 25

Hydrology – Rainfall Runoff relationship in different conditions

Flo

w (

Dis

charg

e)

Time

Natural Surfaces

Densely Urban Surfaces

Private & Confidential 26

Exposure Data Resolution

Other challenges in flood modeling